Bayesian Networks and Their Application to the Reliability of FRP Strengthened Beams

Bayesian Network (BN) is one of the powerful computational methodologies for rare events prediction (i.e., failure probability). Its concept is based on identifying all the different parameters affecting any given problem’s expected output and formulating a network among these variables based on their probabilistic dependencies. Moreover, BNs are capable of handling newly available data and updating the existing model to deliver an updated prediction based on the added information. This study explores the suitability of BNs in predicting the structural reliability of fiber-reinforced polymers externally bonded reinforced concrete beams by estimating their probability of failure. The present article is a preliminary study of a more extensive program to apply the BN framework to assess the structural reliability of bridge systems. The BN will identify the bridge components and explore the probability of failure for each one of them separately. The directed acyclic graph will also illustrate the variables’ dependencies and their conditional probabilities. Finally, the study will evaluate the conditional probability of each specific event occurrence. The BN can also be used for bridge system probability of failure prediction, structural systems monitoring, and maintenance requirements.